In this paper, the formation of the past tense in three varieties of spoken Transcarpathian Rusyn (Lemko, Prešov Rusyn, Zakarpattia Rusyn) is investigated. The study has two aims: First, to quantitatively analyze the areal distribution of past tense forms with and without auxiliaries and to discuss the role of pertinent linguistic and extra-linguistic variables, above all the role of the respective umbrella languages, i.e. Polish, Slovak, and Ukrainian. Second, to demonstrate the possibilities and limits of modern statistical methods such as random forests, conditional inference trees, or mixed-effects binomial logistic regression models that are not yet widely established in Slavic linguistics. The analysis is conducted using the Corpus of Spoken Rusyn (www.russinisch.unifreiburg.de/corpus) and the statistics tool R. It shows that speakers of Lemko prefer to produce past tense forms with the short form of the auxiliary (such as in читала-м), while speakers of Prešov Rusyn favor the long form (читала єм) and speakers of Zakarpattia Rusyn tend to avoid auxiliaries, which complies with the situation in the respective umbrella languages. This means that so-called border effects between the three varieties investigated can be detected. The aforementioned complex statistical methods provide evidence for contact-induced language change: Younger speakers tend to comply with the respective umbrella language even more than older speakers, and the Rusyn varieties thus gradually diverge.

The article is written in German.